摘要
该文提出了一种递推的部分最小二乘算法。该算法利用遗忘因子法对旧参数矩阵进行加权,再和新数据矩阵组合,构成输入输出的数据矩阵,然后利用PLS核算法得到模型。结合实际工业过程的要求,给出了该算法的步骤,并运用该算法建立了煤气化炉合成气组分浓度的软测量模型。结果表明,递推的部分最小二乘算法可以有效的更新模型,使模型适应过程的变化,并能提高模型的预测性能。
In this pap4r, reeursive partial least squares （RPLS） algorithm is proposed. This algorithm uses forgetting factor methodlto be weighted bid parameter matrices, and combines with new data matrix, thus input and output inata matrix is constituted, and the. gets model by using PLS method. Combining with actual indus trial processt：equirementS, the stepsOf this algorithm are proposed, and the paper uses the algorithm to estab- lisll coal gasifier syngas icomponeat concentration of soft--sensor model. Results show that recursive partial least squares algorithm carl effectively update the model, make the model adapt to the changes of the process, and also can improve the prediction performance.
出处
《杭州电子科技大学学报（自然科学版）》
2011年第4期181-184,共4页
Journal of Hangzhou Dianzi University：Natural Sciences
关键词
递推部分最小二乘
煤气化炉
软测量
reeursive partial least squares
coal gasifier
soft-sensor
